Compressive performance of 3D-printed lightweight structures: Infill pattern optimization via Multiple-Criteria Decision Analysis method
Cristina Vălean, Emanoil Linul, Dipen Kumar Rajak
Abstract
• The compressive behavior of MEX-printed PLA parts is similar to cellular materials. • Dimensional accuracy, mass errors and printing time are strongly dependent on infill pattern. • The infill pattern type dictated the brittle or ductile behavior of the 3D-printed parts. • The Star infill pattern highlighted the greatest specific (E, strength and SEA) properties. • The optimization process based on MCDA method identified the Star infill pattern as ideal. The present work addresses a comprehensive compression characterization of Polylactic Acid (PLA) parts produced through the material extrusion (MEX) process, focusing on large strain ranges (0-80%). Additionally, an optimization study of the infill pattern (IP) was conducted using the Multiple-Criteria Decision Analysis (MCDA) method. The study showed that the compressive behavior of the examined parts is similar to cellular materials (e.g., foams) and varies based on the IP, exhibiting brittle, quasi-brittle, or ductile fracture modes. The highest compressive modulus (745.35 MPa) and compressive strength (33.95 MPa) were observed for the Hilbert Curve IP, while the lowest values were recorded for the Lightning IP. Honeycomb IP presented the best energy absorption performance (16.65 MJ/m 3 ), whereas Lightning IP had the lowest (0.14 MJ/m 3 ). In terms of specific properties, Star IP was found as the most efficient, owing to its lightweight design, outperforming other IPs. Considering conventional and specific compressive properties through the MCDA method, Star IP was identified as the optimal pattern across three key criteria: “stiffness-driven”, “strength-driven”, and “energy absorption-driven” structures. Lightning IP exhibited the shortest printing time and lightest weight, whereas Honeycomb IP exhibited the maximum time and weight. Finally, the dimensional accuracy of all MEX-printed specimens was exceptionally high, with dimensional relative errors below 0.15%.